SAN FRANCISCO—In trials of first-line chemotherapy for metastatic colorectal cancer, gains in progression-free survival (PFS) are a good surrogate endpoint for gains in overall survival (OS), according to results of a study presented by Patricia A. Tang, MD, at the 2006 Gastrointestinal Cancers Symposium (abstract 226).
To assess the efficacy of chemotherapy for metastatic colorectal cancer, phase III trials have traditionally used median overall survival, although they have measured a variety of shorter-term surrogate endpoints as well, said Dr. Tang, a drug development fellow at the Princess Margaret Hospital, Toronto.
"Surrogate endpoints such as progression-free survival, time to progression, and response rate have the potential to reduce cost, accelerate drug development, and most important, make promising new therapies available to patients more quickly," she noted.
Using a literature-based analysis, Dr. Tang's research team studied associations of surrogate endpoints with overall survival in this context. They performed a systematic search, identifying 39 randomized controlled trials of first-line chemotherapy in patients with metastatic colorectal cancer for which results were reported between 1990 and 2005. The trials had 87 treatment arms in all and included 18,668 patients.A Strong Correlation
The analyses showed a strong correlation between progression-free survival and median overall survival (r = 0.71, P < .001); a weak correlation between time to progression and median overall survival (r = 0.32, P = .059); and a moderate correlation between response rate and median overall survival (r = 0.54, P < .001), Dr. Tang reported.
When median overall survival was plotted as a function of progression-free survival for each trial arm, "newer agents such as oxaliplatin(Drug information on oxaliplatin) [Eloxatin], bevacizumab(Drug information on bevacizumab) [Avastin], and irinotecan(Drug information on irinotecan) [Camptosar] follow a similar trend to the older drugs, such as 5-FU," she noted, while cautioning that such a correlation does not necessarily imply a causal relationship. Some variation was evident between arms using the same regimen. "Part of that may be due to differences in prognostic factors between trial populations," she commented.
To further test the associations, Dr. Tang and her colleagues studied the correlation between changes in endpoints, comparing the experimental arm with the control arm. "There is a strong relationship between improvements in progression-free survival and improvements in overall survival," she noted (r = 0.78, P < .001). In addition, there were moderate positive correlations between changes in time to progression and changes in median overall survival (r = 0.52, P = .01) and between changes in response rate and changes in median overall survival (r = 0.50, P < .001).
